- “Tay is designed to engage and entertain people where they connect with each other online through casual and playful conversation,” Microsoft said. “The more you chat with Tay the smarter she gets.”
But it appeared on Thursday that Tay’s conversation extended to racist, inflammatory and political statements. Her Twitter conversations have so far reinforced the so-called Godwin’s law – that as an online discussion goes on, the probability of a comparison involving the Nazis or Hitler approaches – with Tay having been encouraged to repeat variations on “Hitler was right” as well as “9/11 was an inside job”.
One Twitter user has also spent time teaching Tay about Donald Trump’s immigration plans.
"Hey, so Markov chains are hip, we should launch one on Twitter."
"Are you sure that will be good for brand?"
"All the kids are on twitter, and all the kids love chatting. And look at the hilarious results from /r/subredditsimulator."
"You're talking about Reddit again, aren't you? Didn't I tell you not to talk about Reddit?"
"But sir, our market research indicates that Cleverbot hits all the demographics we're targeting. And besides, we'll have some human handlers around. Sort of."
- The company launched a verified Twitter account for “Tay” – billed as its “AI fam from the internet that’s got zero chill” – early on Wednesday.
This is dumb. Microsoft is dumb. The people who spent their time last night on twitter responding to a piece of software are dumb. I wish it was 1916.
I like the way you think.
The only problem being MS proved time and time again they are really bad at launching stuff.. W8, W10, MS shop... and all their gaming platform. With their head up on the competition, their history of monopoly and their power, they're doing a really bad job with their result since at least a decade.
So even if they fucked up on purpose, it probably wont serve them.
Not really their play. For one thing, you don't position an asset as an "AI fam from the internet that’s got zero chill” and then act impossibly un-hip. For another, you don't demonstrate your incompetence with AI when your bread and butter is enterprise software. For another, they built a larval Instagram account, a larval Facebook presence and a larval Snapchat presence with no seasoning, no content, and no SEO. I mean... look at this mess.
This looks exactly like what the company that brought you the Zune would do in order to reach those precious "18-24 year old(s)." A company whose search engine uses Google would make a blunder exactly like this.
I remember reading recently (after Google's DeepMind beat Lee Sedol) some people saying that language AI is virtually a solved problem. Anyone who has used Google translate can tell you - even with their resources - it's really far from ideal.
Perhaps the Microsoft team responsible believe that too. Seems they didn't learn anything from those smart-asses at IBM.